Protein synthesis requires the coordination between the ribosome and multiple translation factors to convert the genetic information encoded by the messenger RNA (mRNA) to a polypeptide sequence[1
]. The ribosome is a 2.4 MDa RNA-protein enzyme comprised of two subunits (30S and 50S in prokaryotes) [2
]. After successful initiation, i.e. factor-guided 70S assembly of the two subunits[5
], the ribosome commits to the elongation phase. In each cycle of elongation, the ribosome selects the correct aminoacyl transfer RNA (tRNA) specified by the mRNA codon to the aminoacyl-tRNA binding site (A site)[7
]. Formation of a peptide bond with the peptidyl-tRNA in the adjacent P site transfers the elongating polypeptide from the P-site tRNA to the A-site tRNA. During the translocation step, catalyzed by elongation factor EF-G[9
], the A- and P-site tRNAs must be moved by distances of 20 Å or more to the P site and E site (exit site) respectively[10
], accompanied by the movement of the mRNA by precisely one codon[11
] with respect to the ribosome to maintain the correct reading frame. The E-site tRNA then dissociates spontaneously[1
], leaving the ribosome with a vacant A site and E site, ready for the next round of elongation.
After peptide bond formation ribosome is capable of slowly undergoing spontaneous translocation (at ~3 h−1
]. However EF-G greatly accelerates the process. The overall rate of EF-G dependent translocation in vitro
was measured in bulk to be approximately 20 s−1
at 1 μM EF-G [9
]. This matches well with the rate of elongation in vivo
which is found to be ~ 20 amino acids per second [14
]. After translocation, EF-G dissociates within 50 ms, resetting the ribosome for another round of elongation. These rates frame the timescale for molecular events in translocation.
Determining the molecular mechanism of translocation and open reading-frame maintenance remains one of the key problems in translation. More than four decades ago, a model was proposed by Spirin in which ribosome translocation is controlled by a series of “locking” and “unlocking” events that separates two global conformational states correspondingly termed “unlocked” and “locked” [15
]. In this original model, the ribosomal subunits are tightly associated before peptide bond formation to facilitate manipulation of the tRNA and preserve reading frame on the mRNA. After peptide bond formation, the ribosome unlocks. In this unlocked state, the ribosomal subunits and tRNA can move more freely, facilitating translocation of the tRNA and stepping to the next codon of the mRNA[8
]. Full translocation returns the ribosome to the locked state, once again restricting the motion of the ribosomal subunits and tRNA.
Consistent with this model, cryoelectron microscopy (cryo-EM) and x-ray crystallography structures revealed that the two subunits of the ribosome undergo dramatic conformational changes after peptide bond formation, by which the small (30S) subunit is rotated ~3-10° counterclockwise with respect to the large (50S) subunit[17
]. This rotational movement is possible because the intersubunit contacts, which consistent mainly of RNA-RNA interactions, are relatively labile and can rearrange with little energy cost[18
]. However, it is still unknown how these intersubunit rotations relate to Spirin’s locking and unlocking mechanism. In addition to the intersubunit rotation, there is also a nearly orthogonal rotation of the head domain of the 30S subunit that seems to play a role in controlling the position of tRNAs within the ribosome [20
]. There is thus more than one rotational movement of the ribosome. Recent crystallographic structures of the 70S ribosomal particle have revealed multiple conformational intermediates [22
], and here we call the collection of these stepwise conformational changes leading to translocation “ratcheting.” Aside from these global rotational movements of the ribosome, multiple local conformational rearrangements also occur during the various steps of elongation. For example, the L1 stalk is thought to facilitate the movement of tRNA from the P site to the E site [23
], fluctuating between three distinct conformational states.
Upon peptide bond formation, not only does the ribosome itself undergo dramatic conformational changes prior to translocation, the tRNAs also fluctuate between multiple states. tRNAs can fluctuate freely between the classical (A/A and P/P) state and the hybrid state (A/P and P/E), facilitating the upcoming translocation step catalyzed by EF-G[8
]. These conformational changes are possibly driven by one of the many ribosome conformational rearrangements. Chemical probing, subsequent cryo-EM structures, and single-molecule techniques[7
] all identified a hybrid tRNA configuration, in which the acceptor stems of the A- and P-site tRNAs interact with the P and E sites of the large subunit, respectively, while the anticodon stem loops of the tRNAs remain in the A and P sites of the small subunit. There have been reports of additional hybrid-state intermediates [28
]. This hybrid tRNA configuration can be seen as an intermediate step in translocation. The fluctuation of tRNAs upon peptide bond formation echoes the “unlocking/locking” model proposed by Spirin.
Conformational rearrangements of the ribosome are likely coupled to the tRNA fluctuations and the eventual translocation of the tRNAs and mRNA. Although it is generally assumed that mRNA motion and tRNA movement are coupled mechanically and temporally, this assumption has never been validated directly [22
]. However the link between intersubunit rotation and translocation has been established. If two subunits are cross-linked such that the intersubunit rotational movement is not possible, translocation does not occur[29
]. While ribosome can undergo factor free translocation, EF-G and GTP hydrolysis by EF-G greatly accelerates the rate of translocation (see above) possibly by driving one of these conformational changes. Determining molecular mechanisms that link ribosome conformation and ligand dynamics to EF-G-catalyzed translocation remains a key question in translation.
In this review, we address recent advances in understanding the mechanism of translocation through dynamic studies by fluorescence methods, structural results by cryo-electron microscopy and x-ray crystallography, and computation by molecular dynamics. We incorporate these recent resultswith the current views to formulate a consistent model of translocation.
Real-time dynamics of translocation
Fluorescence approaches, either through single-molecule methods or stopped-flow techniques, probe the dynamics of biological processes with high sensitivity and time resolution. Moreover, fluorescence resonance energy transfer (FRET) provides a method to measure dynamic changes in protein conformation by probing the distance (or potentially the orientation) between a donor dye and acceptor dye (usually separated by 20-80 Å)[30
]. With single-molecule fluorescence, dynamics can be observed directly in real time. Despite the nanosecond timescale of fluorescence excitation and emission, the time resolution in single-molecule fluorescence spectroscopy is frequently limited to tens of milliseconds by current camera technologies so that a sufficient number of photons can be integrated. Bulk fluorescence methods, like quench-flow and stopped-flow, provide high time resolutions of kinetic processes (μs-ms timescales) but suffer from the disadvantage that signals average the behavior of asynchronous ribosomes. Thus bulk fluorescence is limited to investigation of single turnovers, making them unsuitable for observing multiple rounds of translation elongation directly in real-time. Both bulk and single-molecule fluorescence methods have provided a way to track ribosome conformational changes during translocation[14
Aitken and Puglisi (2010) followed the intersubunit rotation of the ribosome in real-time through multiple rounds of elongation by employing single molecule FRET with two dyes on helix 44 of the 30S subunit and helix 101 of the 50S subunit[31
] (). The position of the dyes tracks the rotational state of the 30S subunit body domain[32
]. The intersubunit FRET signal alternates between a high FRET state and a low FRET state on an elongating ribosome (). The elongation cycle begins with ribosome in non-rotated, high FRET state. The transition from the high to the low FRET state occurs upon peptide-bond formation, and corresponds to the rotation of the 30S body by ~ 6 Å. The subsequent reverse transition (back rotation) from the low FRET state back to the high FRET state is catalyzed by EF-G. Neither FRET transition occurs spontaneously, thus the ribosome needs the free energies of peptidyl transfer and GTP hydrolysis during elongation to facilitate transition between these two global states. The existence of two global conformational states separated by large energy barriers as identified by the intersubunit FRET signal is consistent with Spirin’s original model of translation, where in the non-rotated state ribosome is globally locked. Upon peptide bond formation the ribosome rotates and unlocks to permit dynamic tRNA fluctuations between the classic and hybrid states. This puts ribosome in a highly dynamic state in preparation for translocation[33
]. The second transition catalyzed by EF-G likely corresponds to the re-locking of the ribosome upon translocation, where the local conformational dynamics of the ribosome are suppressed [31
Single-molecule translation assays
In the unlocked state, the dynamic fluctuations go beyond tRNAs and involve other parts of the ribosome. For example, Cornish et al
., using FRET between ribosomal proteins S6, S11, and L9, also probed the rotational state of the 30S head domain (though the author did not explicitly mention, it is likely they are observing the 30S head
domain rotation based on crystal structures [22
]) and showed a dynamic equilibrium between the two rotational states of the ribosome[35
] (). This is in contrast with the rotational 30S body
movement observed by the Puglisi group. Interestingly, rather than observing defined chemical steps controlling the conformational changes of the ribosome, Cornish et al
. observed spontaneous rotations of the 30S head domain upon peptide bond formation (). The frequency of the head rotation was dependent on tRNA position in the ribosome and the identity of the bound tRNA, suggesting that conformational rearrangements of the 30S head and tRNAs are linked.
Other single molecule studies have shown the L1 stalk in dynamic fluctuation. Fei et al.
] and Munro et al
. (2010) [36
] used single-molecule FRET between the P-site tRNA and L1 to demonstrate that, prior to peptide bond formation, FRET between L1 and tRNA is stable. Upon peptide bond formation, FRET intensity spontaneously fluctuates. This was interpreted as fluctuations of L1 stalk, however since both L1 and tRNA are mobile these FRET changes do not distinguish between tRNA and L1 mobility. Subsequent studies by Fei et al.
] using FRET between L1 and L9 also demonstrated L1 fluctuations in pre-translocation complexes, confirming that the original experiments were indeed reporting on L1 dynamics.
These results together with a plethora of structural data demonstrate that upon peptide bond formation, ribosomal subunits rotate relative to each other. In agreement with the original “locking and unlocking” model, the ribosome in the rotated state has greater conformational flexibility as seen in the dynamic fluctuations of the tRNA acceptor stems, the 30S head domain, and the 50S L1 stalk. This ribosomal state is often termed unlocked in the literature. However as indicated by the extremely slow rates of factor free translocation [9
], mRNA remains tightly bound to the ribosome. Releasing mRNA to move one codon together with the anticodon loop of the tRNAs requires EF-G binding and/or GTP hydrolysis by EF-G [9
]. This event is also called “unlocking” and adds to the confusion in the literature. Therefore, there are two different uses of the terms unlocking and relocking. The first unlocking happens upon peptide bond formation and permits movements of acceptor stems of the tRNAs while keeping mRNA position tightly locked. The second unlocking occurs upon EF-G binding thus allowing the mRNA transition and movement of the anticodon loops of tRNA. In this model, the pre-translocational ribosome, while conformationally flexible, remains tightly associated to the mRNA as it awaits EF-G arrival. A recent study using single-molecule force microscopy tracking the movement ribosomes on the mRNA with highly sensitive optical tweezers reported that the actual movement of the ribosome on the mRNA occurs quickly, accounting only for a very small fraction of the total time spent in an elongation cycle [38
]. This observation strongly supports the hypothesis that mRNA movement is allowed only during a short period with EF-G catalyzed translocation to insure reading-frame maintenance.
Munro et al.
(2010) examined how EF-G dynamics are correlated with L1 stalk fluctuations[39
]. The authors showed that ribosome L1 stalk closure, as probed by the L1 stalk-tRNA FRET, is correlated with EF-G ribosome interactions. The authors described the closure of the L1 stalk as “unlocking,” based on structural modeling that constraints the subunits to ratchet upon L1 stalk closure. In contrast, the “unlocking” as described by the Puglisi group or Spirin would correspond to the first unlocking step while conformational transitions observed by Munro et al.
likely represent a part of the second unlocking step associated with EF-G. Furthermore, the authors reported an increase of fluorescence as the tRNA is moved from the A site to the P site. The authors found that their measured rate of translocation closely approximates the intrinsic rate at which the ribosome spontaneously closes the L1 stalk, so they argued that EF-G does not measurably accelerate the rate at which this key intermediate in translocation is achieved. Although GTP hydrolysis does accelerate the overall rate of translocation, they suggested that GTP likely plays a role in enabling EF-G to achieve a high-affinity interaction with the ribosome. Thus, their intermediate state is stabilized by EF-G•GTP, but its rate of formation is unaffected by EF-G•GTP. It remains unclear if L1 stalk closure is an upstream event prior to the EF-G driven unlocking step or itself unlocks the mRNA movement per se
In a bulk FRET experiment, Ermolenko el al
. (2011) specifically labeled ribosomal proteins S6, S11, and L9 (the same labeling sites as Cornish et al.
]) for bulk fluorescence experiments[40
]. These FRET probes were positioned to detect the rotational movements of the ribosome subunits, proposed to produce an increase in S11-L9 FRET and an anti-correlated decrease in S6-L9 FRET resulting from a counterclockwise rotation of the 30S subunit (possibly rotation of the head domain) (). Through the use of antibiotics that slow translocation (spectinomycin and hygromycin), Ermolenko et al.
were able to resolve the fast counterclockwise rotation of the two subunits, followed by a slow clockwise reverse rotation. By measuring the quenching of pyrene-labeled mRNA fluorescence, the authors showed that mRNA translocation occurs on the same timescale as the clockwise rotation back from the rotated state to the original state, thereby concluding that the two processes are coupled. Thus, the first counterclockwise rotation would correspond to the mRNA unlocking step in our model. The authors further used EF-G•GDPNP (a nonhydrolyzable GTP analog) or EF-G•GTP with fusidic acid (an antibiotic that inhibits EF-G release after GTP hydrolysis) to demonstrate that the subsequent clockwise rotation does not require EF-G release or GTP hydrolysis. The translocation rate in the presence of GDPNP is slowed down by a factor of ~2.5. However, this clearly is in contrast with the 50-fold inhibition of translocation rate reported by Rodnina et al
] and Munro et al
These dynamic studies revealed that there are multiple locking and unlocking events occurring during an elongation cycle, and they are associated with distinct conformational changes. These conformational transitions are differentially affected by EF-G binding, dissociation, and GTP hydrolysis. Unfortunately, the ambiguous use of the terms “unlocking” and “unlocked state” by the different groups creates unnecessary confusion. However this could be resolved by carefully identifying what conformational movements are tracked and by explicitly specifying what molecular transitions become unlocked or locked.
Structural Insights into Translocation
Despite the power of fluorescence methods to illuminate dynamics, they provide sparse structural information. Cryo-electron microscopy has been used extensively in structural studies of the ribosome, and acts as a potent complement to FRET. Time-resolved and multiparticle cryo-electron microscopy (cryoEM) have revealed of how structural rearrangements on different length scales act concertedly to facilitate mechanical processes central to translation, such as translocation. Despite its relatively lower resolution, cryo-EM has advantages over crystallography because conformationally-diverse substates can be identified within an ensemble of molecules through a variety of existing and emerging particle-classification algorithms. The relative populations of these substates allow assessment of the energetics of their interconversion, provided that they are of sufficiently low energy to appear as detectable populations in the EM data. Thus cryoEM can probe the overall energy landscape of translocation.
Fischer et al.
(2010) employed a time-resolved approach to study thermally-driven ribosomal back-translocation, and identified a large number of ribosomal pre- and post-translocation configurational substates for the E. coli
70S ribosome in the absence of EF-G[41
]. Analysis of the time-dependence of the relative substate populations indicated that substates on each side of the translocation event are in rapid equilibrium, and that the movement of tRNAs from the A/P, P/E hybrid state to the P/P, E/E state itself fully limits the observed rate of translocation. Movement of tRNAs through the ribosome was linked to the formation of tRNA-ribosome contacts, including with the L1 stalk, helix 69 of the 23S rRNA, and the L5 protein. In general, the tRNA motion was shown to be mediated by the sequential formation and dissolution of these tRNA-ribosome contacts, in the context of ongoing 30S body rotation. For each substate of tRNA movement, the authors observed a nearly continuous 30S body rotation and 30S head movements, suggesting a coupling of tRNA movement and global changes in ribosome conformation (). The transition from Pre1 to Pre2 (likely peptide bond formation) results in a large 30S body rotation counterclockwise, consistent with single-molecule FRET measurements discussed above. Up until translocation, the 30S head also rotates. After translocation (from Pre5 to Post1), both the 30S head and 30S body rotates back, resetting the ribosome for the next round of elongation.
Cryo-EM reveals sub-states of translation
A similar paradigm for ribosome dynamics was also uncovered in the cryo-EM study of Ratje et al.
(2010) on Thermus thermophilus
70S ribosomes bound to EF-G in the presence of fusidic acid, which results in EF-G catalyzed GTP hydrolysis, but not the subsequent translocationally-relevant conformational changes in, or dissociation of, EF-G•GDP[42
]. Results from this study also revealed a number of configurations of the pre-translocation ribosome that were intermediates on the translocation pathway. In particular this study identified an “intra-subunit” hybrid state, termed a pe/E state, where the tRNA made contacts with both the P-site components of the 30S head and its E-site platform components at the same time. The data led to the proposal that EF-G binds to the pre-translocation ribosome, stabilizing the fully “ratcheted” state, before translocation occurs in a manner that is facilitated by GTP hydrolysis. GTP hydrolysis was proposed to lead to reverse-ratcheting of the 30S head and body, which drives tRNA motion. As the tRNAs then move through the hybrid ap/P and pe/E states into the classical P/P and E/E states, interaction of Domain IV of EF-G is proposed to decouple back-ratcheting from reverse tRNA motion, providing the net directionality. Molecular dynamics and other studies also implicated Domain IV as playing a central role in translocation (see below). This cryo-EM study also implicated a swiveling motion of the 30S head as being instrumental in guiding tRNA transit.
These combined results highlight the importance of multiple ribosome conformational dynamics in guiding tRNA and mRNA motion, and show that contacts between the ribosome and tRNAs facilitate an energetically inexpensive progression through a large number of intermediates states during translocation. The two studies are consistent with the model proposed by an earlier crystallography study by Zhang et al.
], showing that there are multiple conformational intermediates during elongation. In that study, Zhang observed at least four intermediates of ratcheting: ratcheting likely begins with the 30S subunit body rotation, continuing with the 30S platform and head domains, and completes with rearrangement of the central bridges. It is clear that in addition to the rotational movement of the 30S subunit, other large-scale conformational changes, like the 30S head movement, are essential for translocation. High-resolution structures of the ribosome in different rotated states with intact tRNAs will be required to complete our molecular understanding of the large-scale conformational rearrangements in the ribosome that allow ratcheting and translocation.
Linking structural and dynamic information
Structural snapshots from X-ray crystallography and cryo-EM suggest that many conformational rearrangements of the ribosome, tRNAs, and EF-G must occur during translocation. Nevertheless, high-resolution structures of the transition states in between those snapshots are not available, due to the intrinsic high energy and short lifetimes of these states. Cryo-EM imaging techniques have been successful in capturing some of the intermediate states, albeit at lower resolution and sometimes with missing components. Moreover, the short time-scales (microseconds or shorter) involved in some highly energetic states along the reaction coordinate for translation are simply inaccessible to most methods. Computational approaches provide a link between structure and dynamics. While the size of the ribosome (~3.2 million atoms) and the long time scales (up to 100 ms) involved in translocation [43
] render all-atom molecular dynamics (MD) simulations of the complete process intractable, shorter simulations on parts of the elongation cycle starting from high-resolution structures have probed important structural and dynamic features of the transition states.
Simulations of the whole ribosome remain challenging, considering the large number of atoms and need for robust potential energy functions; however, recent advances are exciting. Li et al
] simulated structural dynamics of the large conformational changes of EF-G on a ribosome to drive translocation. Their results suggest that EF-G has a large degree of conformational flexibility, especially of Domain IV (the elongated domain that reaches into the A site of the ribosome) with respect to the rest of the molecule, and this flexibility likely lies at the center of the function of EF-G during translocation (). The conformational variability of EF-G when bound to the ribosome could be driven by interactions of EF-G with the GTPase-associated center of the large subunit, specifically the L11 lobe, as the ribosome changes its ratcheting state.
Recent achievements in all-atom MD simulations are promising for dissecting the structural and dynamic features of short-lived transition states of the ribosome complex with various elongation factors during translocation. MD provides a great complement to cryo-EM studies with low spatial and time resolution and the dynamic fluorescence studies with higher time resolution but little spatial resolution. While improvements in computational techniques will make MD an essential tool for understanding ribosomal function, observations originating from MD simulations must be experimentally tested. The integration of computation and experiment is a challenge for upcoming years.